DSD Windows Port

Status
Not open for further replies.
Joined
Jan 28, 2013
Messages
56
Location
Hanmer,Ontario,CA
Provoice screenshot

Judging by the screen shot what's missing? Poor signal?... I get some good clear dispatch (not mobile)conversations, then when dispatch keys up again I get choppy\broken conversation. Using the the SDR1172 piped throught VC to DSD1.6..
Any sugestions?


I've played with all the filters, inlvl up 30 to 90, VC or VB cable settings....it all sounds the same,,,
 

Attachments

  • screen.jpg
    screen.jpg
    229.1 KB · Views: 1,741
Last edited:

mk262

Member
Premium Subscriber
Joined
Sep 17, 2010
Messages
454
Location
Flagler County
Judging by the screen shot what's missing? Poor signal... I get some good clear dispatch (not mobile)conversations, then when dispatch keys up again I get choppy\broken conversation. Using the the SDR1172 piped throught VC to DSD1.6..
Any sugestions?

need sdr# screenshot
 
Joined
Jan 28, 2013
Messages
56
Location
Hanmer,Ontario,CA
sdr screenshot

I keep the resolution at 512 to keep cpu usage low. Front end sampling rate 0.25 to get the smallest freq. window (also reduces cpu usage)

When all said and done, cpu peaks at 50% to 70% when getting good conversations.

I've also played with latency from 100 to max 2000, with no effect.
 

Attachments

  • screen2.jpg
    screen2.jpg
    125 KB · Views: 1,682
Joined
Jan 28, 2013
Messages
56
Location
Hanmer,Ontario,CA
i'v tried it,

Your audio level is too low, it shows inlvl: 1% to 6%. Needs to be up around 25% or so

I've tried at all levels but I will try again.

I get good decode when the level stays at the constant input,,,but with some transmissions that I get are garbled the inlvl is way high.

would this indicate a poor signal with lots of distortion cause the inlvl to shoot up?
 
Last edited:
Joined
Jan 28, 2013
Messages
56
Location
Hanmer,Ontario,CA
Cracking sound then garbled voice

The sound of a encrypted message is very noticeable.

Eureka!! I finally tried the setup on my desktop.. I thought they were identical with respect to cpu power but made a mistake. One is duo core and the other is quad core...I guess multi threading capabilities. So dsd needs lots of cpu power to operate, "to bad they never said system requirements"

I guess a high pitch cracking sound with the voice garbled is a indicator of a struggling dsd.
 

groovy

Member
Joined
Jul 10, 2013
Messages
29
Were the two computers using the identical build of dsd?

I think there are huge opportunities for code optimizations in dsd/mbelib. The 3600 codec was recently optimized to use bit shifting when setting up the parameters, and the same could be applied to to the 7200bps codec in p25. But that is probably a 1-2% gain. IMHO The bigger gain is to either avoid or replace with tables some of the floating point operations in mbelib.c- specifically the cosine and square root calculations. I haven't profiled the code, but that would be my impression just quickly looking at it.

The sound of a encrypted message is very noticeable.

Eureka!! I finally tried the setup on my desktop.. I thought they were identical with respect to cpu power but made a mistake. One is duo core and the other is quad core...I guess multi threading capabilities. So dsd needs lots of cpu power to operate, "to bad they never said system requirements"

I guess a high pitch cracking sound with the voice garbled is a indicator of a struggling dsd.
 

EricCottrell

Member
Premium Subscriber
Joined
Nov 8, 2002
Messages
2,414
Location
Boston, Ma
Hello,

Some interesting observations earlier in the thread. I have a nearby EDACS Extended Addressing (EA) system using ProVoice and would be interested in decoding more of the data.

I am not surprised that the automute does not work well. I noticed the NAC and other display values vary during decode, even with a good signal. One reason P25 decode is fair to poor is the P25 moving-average filter is missing. This filter allows a FM demodulator to decode both C4FM and QPSK. Another problem is the P25 bit sync has too much jitter because of the simple sync detector.

Adding the proper filtering to the existing code causes problems with mode autodetection and the other modes. One way to do it properly is to split the incoming audio stream into multiple streams. Each stream would go through the proper filter and then to the proper sync detection and decoder for each similar mode. The upside is the program could autodetect all the modes. The downside is this may take more processor time and require restructuring of DSD code. The processor time increase may be mitigated by the elimination of the unneeded QPSK code.

DSD does not implement error correction, only mbelib does for voice. I think the voice error correction corrects much of the dsd code shortcomings. When I implemented table driven error correction for DMR data, I found weak signal decoding of data noticeably improved over just having the DMR filter. I have also added code to decode most of the DMR CSBK and TLC information, so I use dsd instead of DMRDecode.

There are a bunch of other annoyances. The dsd code is structured to only decode one DMR voice slot when it is present, so control channel decoding goes away. When two voice slots are active, it will decode one voice slot and then jump to the other voice slot when TLC frames appear on the first voice slot between transmissions.

I have a feeling that most of the dsd code would need to be restructured to fix all the problems, or it may be easier to just start from the ground up. Apparently this was done with DSDplus, although I would not use it. I can not add DMR decoding, and USB3000 support since DSDPlus is closed source.

I do not want to bring another dsd clone into the world. I am more interested in developing a program that decodes digital trunked radio systems, which means a new program from the ground up. I am using dsd to test some concepts. This development will take months.

Another improvement I am working on is to use ncurses for a better display. I think adding this may break a cygwin windows port as it is not clear if cygwin can use the ncurses library.

73 Eric
 

landry76560

Newbie
Joined
Mar 30, 2015
Messages
4
how to compile and dsd mbelib-master-master.

hello,


here I would like to run speesh digital decoder v1.7 windows vista with no experience of

cygwin software to compile and dsd mbelib-master-master and run could you give me the

march forward for this? or did you compile yourself after multiple search on the net has nothing

Giving is why I did call you

waiting for your reply.

Regards,
landry
 
Status
Not open for further replies.
Top